Read the Full Video Transcript

Rahul Aggarwal: Right. Thanks, everyone, and thanks, Tom, Jeremie, Johannes, and Bella, for inviting me to speak today and for organizing this event. I've been listening online for a good chunk of yesterday and today, and it's just been fantastic to see the questions and the engagement. Before I talk about the combination of RLT and immunotherapy, I just want to piggyback on what Amanda said a little bit ago about the multidisciplinary aspects of care for these patients. It's been hugely beneficial for our patients to see both med onc and nuclear medicine with each cycle of RLT at our institution. We've learned a lot from each other.

A real shout-out to Tom and Amanda. They regularly participate in our tumor board meetings. We've learned a lot from them in terms of how to interpret SPECT-CTs and PSMA PETs and how to select the right patients for treatment. We're really, really excited about this new era of collaboration between nuc med and med onc. I'm a GU medical oncologist at UCSF. I'm happy to share some of our data that we have, which is preliminary but encouraging. I think Johannes presented some of that data, but really looking at the combination of RLT and immunotherapy in patients with mCRPC.

Okay. The other way? Hitting the right one. Okay. This concept of triggering or trying to prime in prostate cancer, which we know is immunologically cold and trying to convert it to a hot tumor with more tumor-infiltrating lymphocytes, is obviously not a new concept. We know that external beam radiation therapy combined with immunotherapy has been tested in a lot of tumor types, including prostate cancer. It's a little bit of a double-edged sword.

You do see that there's DNA damage and immunogenic cell death, and we have some encouraging preclinical data, but you also see a lot of unfavorable characteristics. Upregulation of myeloid-derived suppressor cells, increased expression of checkpoint proteins, ablation of preexisting T cell subsets, and a lot of mixed preclinical data. Some studies show there's a positive synergy between external beam radiation and checkpoint inhibition. Other studies show not so much.

We have one prior phase III trial that randomized patients to a round of SBRT, followed by ipilimumab versus placebo, which just missed meeting its overall survival endpoint. But in long-term follow-up, it did look like there was a tail on the Kaplan-Meier curve, and there were excess survivors in the group that received ipilimumab. Perhaps there's a role for the combination of immunotherapy plus radiation treatment in prostate cancer.

What we hypothesized, which I think is still unproven, but really, can targeted RLT, by being a more tumor-specific type of delivery system for radiation, achieve a more favorable balance between the antitumor immunity versus less immunosuppressive effects compared to external beam radiation? That's really what we set out to test. That's really the rationale for the way that we designed our phase II or phase Ib study, which was really a single priming dose of RLT. This is different than the PRINCE study, which used the standard six cycles of lutetium.

Can we give a single cycle to really prime the immune system? By doing that, can we potentially avoid some of the ablative effects of giving repeated doses of radiation? Can we extend the duration of response with pembrolizumab after that single dose of lutetium and really minimize toxicity and set patients up to be able to receive future doses of lutetium, which we have some of that data in our study that I'll show you?

Then, as a follow-on to this initial study, can we use repeated doses of lutetium but using this adaptive dosing, which we've heard about earlier today? Importantly, can we, by combining RLT with pembrolizumab, see responses in lesions that we might not expect to respond well to RLT? One of the important exploratory objectives of our study is really looking and doing a deep dive on PSMA-low or negative lesions. What kind of activity do we see in those? Is there any clinical evidence for antigen spread, which is a really difficult question to answer? We tried to do the best we could out of the data that we had.

This is the design of our phase Ib study, a proof of concept, relatively limited sample size, single-arm study where we had two parts to our trial. This was an mCRPC patient population. We intentionally chose a more liberal selection strategy for patient selection. These patients just had to have three or more positive lesions on PSMA PET. We did not exclude patients who had negative soft tissue lesions according to the so-called VISION criteria. Those patients were actually allowed to enroll. We do have some data on what the outcomes were for that subset of patients. They just had to have had one or more prior ARSIs and couldn't have had prior taxanes for mCRPC. This is a less heavily pretreated population than the group that was in VISION.

We tested three schedules in Part A of the study. This was, albeit with a limited sample size, but really three schedules of the priming dose of lutetium, followed by continuous pembrolizumab until radiographic or clinical PD. Schedule 1 was really giving the priming dose first, followed by pembrolizumab. Schedule 2 really gave concomitantly the lutetium and pembrolizumab on the first day, and then in Schedule 3, actually starting the pembrolizumab first and then giving the lutetium on cycle two, day one. In all patients, treatment was continued with pembrolizumab until radiographic clinical PD. We did the best we could in terms of choosing the right schedule to move forward with. In Part B, we chose Schedule 1, as I'll show you, and really enrolled an additional 25 patients just to gather additional data, both from a safety and preliminary efficacy standpoint.

These are the eligibility criteria for the study. These are mCRPC patients, one or more prior ARSIs, no prior chemo for mCRPC. Importantly, we did require measurable disease by RECIST. This did enrich for a higher percentage of patients with visceral metastases as compared to a typical mCRPC study, as I'll show you. We did utilize a more liberal selection criteria for PSMA PET, and they could not have received a prior checkpoint inhibitor. Importantly, this was a genomically unselected patient population. We did not enrich for microsatellite high patients or those with DNA damage repair mutations, those groups that you might predict would do better with pembrolizumab.

The study endpoints were typical for a phase I study. We looked at safety and recommended phase II schedule. A number of efficacy endpoints, including objective response rate, PSA50 response, duration of response, rPFS, overall survival, and a number of correlative analyses, a number of which are still underway. But I'll show you some of the preliminary data that we've generated so far.

Here are the baseline characteristics. In total, we enrolled 18 patients in Part 1, 25 patients in Part 2. In total, 43 patients. Typical baseline characteristics, you can see that about half of the patients have received both prior abiraterone and enzalutamide. About 15% have received prior taxanes in the hormone-sensitive disease space. We did see 30% of our patients with visceral mets. Again, we required measurable disease by RECIST criteria. This was enriched for a higher-risk population than the typical mCRPC study. Because we didn't pre-specify any genomic selection and we have other trials that do, we inherently enrich for patients that are wild-type with respect to these mutations. These were all MS-stable patients and only three of which had a DNA repair pathway mutation.

These are the characteristics for this study in terms of patient disposition and screen fail rate. I just want to highlight, the PET screen fail rate was lower, as might be expected, because we utilized that more liberal selection strategy. In terms of the patients that were eventually treated, we actually have had some durable responses, including one patient who's been off treatment on a treatment break for a number of months who remains without any evidence of disease on scans.

The safety profile is worth highlighting. We're using only a single dose of lutetium, followed by pembrolizumab. The hematologic toxicity was pretty minimal with this study. As you might expect, this is a pre-chemo population, and we're only giving a single dose of lutetium. There was no grade three or higher hematologic toxicity, really no high-grade dry mouth, dry eyes, or any of the expected effects you'd see with repeated doses of lutetium.

Just to summarize, only 2 out of 43 patients had grade three or higher treatment-related adverse effects, so 5% of patients, only one of which discontinued due to a side effect. There were two irAEs. These are expected with pembrolizumab. The rate was typical. One was pneumonitis. The other was an inflammatory arthritis. The pneumonitis did cause a discontinuation of pembrolizumab, but that patient remained without progression for a number of months. No significant lutetium-related side effects were seen.

As Johannes showed already, this is really pretty impressive in terms of the responses we see. I'll talk about some of the commentary in a minute. But the overall objective response rate, which was one of the key efficacy endpoints, keep in mind, we had measurable disease in all patients, so we're able to measure the ORR pretty carefully. It was 56% in this patient population, including the patients with CR, which you can see here.

More importantly than the objective response rate is really the durability. Is there any preliminary evidence that pembrolizumab is contributing anything? The duration of response of 8.1 months was encouraging, our PFS of 6.9 months median. Then we looked at some of those higher-risk patients, those that had visceral metastases, including in the lung and liver. We know these are patients associated with the worst outcomes. We see a 45% confirmed response rate in that subset of patients.

It didn't appear that there were any significant differences in response rates across the three schedules, as shown in the different colors here, but again, small numbers of patients. The duration of treatment as shown in the swimmer's plot is shown here. Median follow-up of 16.5 months. We had a duration of treatment of about five months, but a real spread. That just highlights this subset of patients that can really have a durable response as seen on the trial.

We really wanted to get at that question of, can we extend the benefit of lutetium beyond those patients with really high uptake? We chose a liberal selection strategy on a lesion-by-lesion basis, trying to really look at what's the lesion-specific response based on the PSMA PET obtained at baseline. Not unexpectedly, there was a correlation between the higher the level of uptake, the more likely it is to see shrinkage of the tumor on scan. That explains the negative correlation.

But importantly, among the lesions with lower PSMA PET uptake, we did see significant shrinkage in a subset of those patients. Of the 43 patients that we enrolled, 7 did have PET scans and CT scans that would have disqualified them for VISION, meaning they had one or more negative soft tissue lesions on CT. Of those, we actually saw a subset that had benefited from the treatment. Small numbers, very preliminary, but indicates maybe there's something to the combination of RLT and immunotherapy.

The PSA waterfall plot shown here, the PSA50 response rate was 56%, PSA90 response was 16%, and PSA-PFS was 3.5 months. One of the things that really set us up for the phase II trial, which I'll show in a minute, was really the lag time between PSA and radiographic PD and this concept of adaptive dosing and seeing this lag between when the PSA rose and when scans rose, not surprising, but allows us the stability to think about adaptive dosing based on serum PSA.

We looked at PET scans both at baseline and progression. We didn't get on-treatment PET scans, just given the limitations of the study design and so forth, but really saw that, as we've seen with other studies, most of the time, PSMA PET expression and uptake are retained at the time of progression. You can see the mean SUV max is there, 28.9 at baseline, 21.7 at progression. We did have a subset of patients who subsequently were treated with lutetium as standard of care at some future time point once lutetium became FDA-approved, the majority of which did have a response, suggesting that you can give repeated doses of lutetium and generally, you see favorable outcomes.

This is just an example of one of the patients that we treated on the trial. It's a patient of mine who had both abiraterone and enzalutamide, a pretty short duration of time to castration-resistance. Really large, bulky soft tissue disease. This is a rib metastasis shown, highlighted there with a pretty bulky soft tissue component. Really headed towards having not very good outcomes. We gave the single dose of lutetium, followed by pembrolizumab.

You can see that, really, for over the course of a year and longer actually, the patient had a really durable partial response on imaging and a very significant decline in PSA, as you can see there. Again, how much is the contribution of the single dose of RLT versus the pembrolizumab? But the durability of responses is definitely intriguing that we saw during the course of the study.

Just talking a little bit about the immune correlative studies. This was led by Larry Fong and the Cancer Immunotherapy Lab. Really, this work is still ongoing. We showed some of it in the Lancet Oncology paper. There's a lot more ongoing. We really intentionally tried to get paired biopsies and required measurable disease in part so we could get those biopsies both before and after the priming dose of lutetium. There's a lot of work ongoing using single-cell RNA-seq and MIBI and some of the more sophisticated techniques to really try to understand at the tumor microenvironment level what the dose of RLT is doing to the tumor microenvironment.

A lot of that is still ongoing, but we did have some whole blood analyses that we included in this paper using CyTOF to really try to understand what's happening in the peripheral compartment with the different immune cell components. Are they going up? Are they going down? What's happening in the responders versus nonresponders?

This is just the gating strategy that's used for CyTOF as part of this study. There are a couple of findings that were pretty interesting that we included in the paper. One, not surprisingly, we're giving pembrolizumab. We see a downregulation of PD-1 really across all three schedules that were analyzed. But more importantly, we're really seeing differences in the responders versus nonresponders. This is just based on peripheral whole blood CyTOF.

I think we're really excited to get some of the follow-up data with the tissue and really see what's happening in the tumor microenvironment. But even when you look in the whole blood, you see differences and you see an upregulation of cytotoxic T cells and downregulation of myelosuppressive cell populations in the objective responders who had an objective response on CT by RECIST compared to those that don't. We're really excited to see the tissue analysis, the single-cell analysis both from the blood and the tissue. That'll be really important as we sort of think about what RLT does and if it's truly providing a priming effect.

Oops, sorry. There are clearly some limitations of this study. This is really just the first foray into using a single priming dose of lutetium and pembrolizumab. The data from this study, as well as the PRINCE study, which I'll show in a second, really, when you're trying to interpret the clinical data, it's hard to know what is the true impact of the IO. We know that pembrolizumab on its own has very limited activity in mCRPC. Ultimately, we'll need randomized trials to really sort that out.

We didn't get PSMA PET imaging on study. That would have been nice to really assess the metabolic response, especially in the bone. We did get CT scans and looked at the soft tissue pretty carefully. Then we only enrolled six patients per schedule. We don't really know what's the right schedule. As Johannes pointed out, preclinical studies may be really important because you have so many variables, dose, timing, schedule to really sort out when you're trying to think about combining RLT with immunotherapy.

There is some preliminary data that, in fact, giving the doses concomitantly and then following up with a checkpoint inhibitor might even be better than giving the RLT first, followed by the checkpoint inhibitor, which in turn may be better than starting the checkpoint inhibitor first. But we just need more data to really sort that out. Then dose, we used the standard dose of lutetium, 7.4 gigabecquerels. There's really no clinical data that would say that's the optimal dose to prime an immune response. A lot more to figure out in terms of how to combine these therapies.

The PRINCE study that Johannes mentioned is really a nice study conducted by Peter Mac and Shahneen Sandhu, and really quite different from our study. It's really a standard six cycles of lutetium, as opposed to one cycle, six cycles, so the standard schedule. This is a more heavily pretreated population. Three-fourths of the patients had prior docetaxel, so more heavily pretreated. Did not require measurable disease, so more predominance of bone-only patients, maybe a little bit more favorable from that standpoint, but really encouraging.

You see a median rPFS of 11.2 months. With all the caveats of cross-trial comparisons, that looks favorable compared to VISION. Obviously, this is a nonrandomized study. As with our study, there is this subset of patients, despite careful selection by PSMA PET, that just seems to be resistant, that's highlighted in the red circle on the waterfall plot, that just don't seem to respond. I don't think we yet have a good handle on why that occurs.

I think there are a lot of unanswered questions. The paired biopsies are going to be important to really understand what's happening in the tumor microenvironment. What's the clinical contribution? We need randomized studies. There are randomized trial data that should be coming that I think are important. We don't really know how to maximize that immunogenic impact. What's the right dose? What's the right schedule? Should we be giving actinium or some sort of alpha particle radiation? Does that achieve a better effect?

We have an ongoing PCF Challenge grant to really try to look, at least preclinically, at the different models, test the different types of radiation, and see really what achieves the best immunogenic impact. Then other IO partners. We chose pembrolizumab because it's in the clinic. We know how to use it. We're comfortable with it. But what about other co-modulatory or co-stimulatory targets? There's promising data with CD28, CTLA-4 blockade, many others.

We think about targeting multiple antigens. We know that mCRPC is very heterogeneous. There is PSMA-low, PSMA-negative disease. In the IO space, we have now encouraging data with targeting other surface antigens, including STEAP1 with some of the bispecific data that came out at ESMO. Could we think about a rational combination approach that not only serves as a priming effect for the immunotherapy but targets multiple antigens at the same time? I think that's really how we're going to move the field forward. There's just too much heterogeneity to think about targeting a single antigen and really achieving durable responses in the majority of patients.

This is our follow-on study to the phase Ib data that I presented. Just want to highlight a couple of things. We're taking the approach of, okay, we saw really encouraging results with our single priming dose, can we do better by utilizing an adaptive design? It's really two variables, adaptive dosing and layering on top the pembrolizumab. We give a lutetium priming dose, we wait until there's PSA progression by working group criteria, and then we give a second dose, up to a total of six doses. Patients still get the full treatment with RLT but spaced out at intervals.

Then we space out the pembrolizumab as well. We give it Q6 weeks, for which there's now good data that that seems to be just as good as Q3 weeks, really trying to minimize patient impact, less frequent visits to the clinic, hopefully less toxicity by spacing out the doses of RLT, looking at rPFS at 12 months as the primary endpoint of the study, using a VISION-type criteria in terms of selecting patients for the trial. We just launched this investigator-initiated trial. We just started to enroll patients in the last month and are really excited about the data we might see.

Still won't be definitive because it's a nonrandomized study, but I think we'll provide that data of, okay, what if we give multiple doses of RLT, integrating the biopsies, which we're going to do again, to not only test prime but reprime. If a patient's PSA starts to go up, we give another dose of RLT. What's happening in the immune microenvironment in the peripheral compartment? Can we reprime or rejuvenate the immune system to attack the cancer? The statistical design is shown here. The 12-month rPFS is the primary endpoint of the study. We use VISION as our benchmark for that. But, really, it's to get more data and, really, with respect to all the efficacy endpoints that we have planned.

With that, I'd just like to conclude. I want to thank patients and families. The phase Ib study, that was really in the midst of the COVID pandemic and asking them to do paired biopsies and all the extra blood draws, really a huge ask. The entire GU medical oncology program, radiology with Tom, the immunotherapy lab with Larry Fong, Dave Oh, and others, and our funding support, including PCF. Really appreciative. Thank you very much.